ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/JSON-XS/XS.xs
(Generate patch)

Comparing JSON-XS/XS.xs (file contents):
Revision 1.38 by root, Mon Jun 11 03:18:07 2007 UTC vs.
Revision 1.56 by root, Thu Jul 26 11:33:35 2007 UTC

1#include "EXTERN.h" 1#include "EXTERN.h"
2#include "perl.h" 2#include "perl.h"
3#include "XSUB.h" 3#include "XSUB.h"
4 4
5#include "assert.h" 5#include <assert.h>
6#include "string.h" 6#include <string.h>
7#include "stdlib.h" 7#include <stdlib.h>
8#include "stdio.h" 8#include <stdio.h>
9#include <float.h>
9 10
10#if defined(__BORLANDC__) || defined(_MSC_VER) 11#if defined(__BORLANDC__) || defined(_MSC_VER)
11# define snprintf _snprintf // C compilers have this in stdio.h 12# define snprintf _snprintf // C compilers have this in stdio.h
12#endif 13#endif
13 14
15// some old perls do not have this, try to make it work, no
16// guarentees, though. if it breaks, you get to keep the pieces.
17#ifndef UTF8_MAXBYTES
18# define UTF8_MAXBYTES 13
19#endif
20
14#define F_ASCII 0x00000001UL 21#define F_ASCII 0x00000001UL
15#define F_LATIN1 0x00000002UL 22#define F_LATIN1 0x00000002UL
16#define F_UTF8 0x00000004UL 23#define F_UTF8 0x00000004UL
17#define F_INDENT 0x00000008UL 24#define F_INDENT 0x00000008UL
18#define F_CANONICAL 0x00000010UL 25#define F_CANONICAL 0x00000010UL
19#define F_SPACE_BEFORE 0x00000020UL 26#define F_SPACE_BEFORE 0x00000020UL
20#define F_SPACE_AFTER 0x00000040UL 27#define F_SPACE_AFTER 0x00000040UL
21#define F_ALLOW_NONREF 0x00000100UL 28#define F_ALLOW_NONREF 0x00000100UL
22#define F_SHRINK 0x00000200UL 29#define F_SHRINK 0x00000200UL
30#define F_ALLOW_BLESSED 0x00000400UL
31#define F_CONV_BLESSED 0x00000800UL
23#define F_MAXDEPTH 0xf8000000UL 32#define F_MAXDEPTH 0xf8000000UL
24#define S_MAXDEPTH 27 33#define S_MAXDEPTH 27
34#define F_MAXSIZE 0x01f00000UL
35#define S_MAXSIZE 20
36#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
25 37
26#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) 38#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
27 39#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
28// F_SELFCONVERT? <=> to_json/toJson
29// F_BLESSED? <=> { $__class__$ => }
30 40
31#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 41#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
32#define F_DEFAULT (9UL << S_MAXDEPTH) 42#define F_DEFAULT (9UL << S_MAXDEPTH)
33 43
34#define INIT_SIZE 32 // initial scalar size to be allocated 44#define INIT_SIZE 32 // initial scalar size to be allocated
48#endif 58#endif
49 59
50#define expect_false(expr) expect ((expr) != 0, 0) 60#define expect_false(expr) expect ((expr) != 0, 0)
51#define expect_true(expr) expect ((expr) != 0, 1) 61#define expect_true(expr) expect ((expr) != 0, 1)
52 62
53static HV *json_stash; // JSON::XS:: 63static HV *json_stash, *json_boolean_stash; // JSON::XS::
64static SV *json_true, *json_false;
65
66typedef struct {
67 U32 flags;
68 SV *cb_object;
69 HV *cb_sk_object;
70} JSON;
54 71
55///////////////////////////////////////////////////////////////////////////// 72/////////////////////////////////////////////////////////////////////////////
56// utility functions 73// utility functions
57 74
58static UV * 75inline void
59SvJSON (SV *sv)
60{
61 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
62 croak ("object is not of type JSON::XS");
63
64 return &SvUVX (SvRV (sv));
65}
66
67static void
68shrink (SV *sv) 76shrink (SV *sv)
69{ 77{
70 sv_utf8_downgrade (sv, 1); 78 sv_utf8_downgrade (sv, 1);
71 if (SvLEN (sv) > SvCUR (sv) + 1) 79 if (SvLEN (sv) > SvCUR (sv) + 1)
72 { 80 {
107typedef struct 115typedef struct
108{ 116{
109 char *cur; // SvPVX (sv) + current output position 117 char *cur; // SvPVX (sv) + current output position
110 char *end; // SvEND (sv) 118 char *end; // SvEND (sv)
111 SV *sv; // result scalar 119 SV *sv; // result scalar
112 U32 flags; // F_* 120 JSON json;
113 U32 indent; // indentation level 121 U32 indent; // indentation level
114 U32 maxdepth; // max. indentation/recursion level 122 U32 maxdepth; // max. indentation/recursion level
115} enc_t; 123} enc_t;
116 124
117inline void 125inline void
178 STRLEN clen; 186 STRLEN clen;
179 UV uch; 187 UV uch;
180 188
181 if (is_utf8) 189 if (is_utf8)
182 { 190 {
183 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
184 uch = decode_utf8 (str, end - str, &clen); 191 uch = decode_utf8 (str, end - str, &clen);
185 if (clen == (STRLEN)-1) 192 if (clen == (STRLEN)-1)
186 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 193 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
187 } 194 }
188 else 195 else
192 } 199 }
193 200
194 if (uch > 0x10FFFFUL) 201 if (uch > 0x10FFFFUL)
195 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); 202 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
196 203
197 if (uch < 0x80 || enc->flags & F_ASCII || (enc->flags & F_LATIN1 && uch > 0xFF)) 204 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
198 { 205 {
199 if (uch > 0xFFFFUL) 206 if (uch > 0xFFFFUL)
200 { 207 {
201 need (enc, len += 11); 208 need (enc, len += 11);
202 sprintf (enc->cur, "\\u%04x\\u%04x", 209 sprintf (enc->cur, "\\u%04x\\u%04x",
216 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 223 *enc->cur++ = hexdigit [(uch >> 0) & 15];
217 } 224 }
218 225
219 str += clen; 226 str += clen;
220 } 227 }
221 else if (enc->flags & F_LATIN1) 228 else if (enc->json.flags & F_LATIN1)
222 { 229 {
223 *enc->cur++ = uch; 230 *enc->cur++ = uch;
224 str += clen; 231 str += clen;
225 } 232 }
226 else if (is_utf8) 233 else if (is_utf8)
247} 254}
248 255
249inline void 256inline void
250encode_indent (enc_t *enc) 257encode_indent (enc_t *enc)
251{ 258{
252 if (enc->flags & F_INDENT) 259 if (enc->json.flags & F_INDENT)
253 { 260 {
254 int spaces = enc->indent * INDENT_STEP; 261 int spaces = enc->indent * INDENT_STEP;
255 262
256 need (enc, spaces); 263 need (enc, spaces);
257 memset (enc->cur, ' ', spaces); 264 memset (enc->cur, ' ', spaces);
267} 274}
268 275
269inline void 276inline void
270encode_nl (enc_t *enc) 277encode_nl (enc_t *enc)
271{ 278{
272 if (enc->flags & F_INDENT) 279 if (enc->json.flags & F_INDENT)
273 { 280 {
274 need (enc, 1); 281 need (enc, 1);
275 encode_ch (enc, '\n'); 282 encode_ch (enc, '\n');
276 } 283 }
277} 284}
279inline void 286inline void
280encode_comma (enc_t *enc) 287encode_comma (enc_t *enc)
281{ 288{
282 encode_ch (enc, ','); 289 encode_ch (enc, ',');
283 290
284 if (enc->flags & F_INDENT) 291 if (enc->json.flags & F_INDENT)
285 encode_nl (enc); 292 encode_nl (enc);
286 else if (enc->flags & F_SPACE_AFTER) 293 else if (enc->json.flags & F_SPACE_AFTER)
287 encode_space (enc); 294 encode_space (enc);
288} 295}
289 296
290static void encode_sv (enc_t *enc, SV *sv); 297static void encode_sv (enc_t *enc, SV *sv);
291 298
300 encode_ch (enc, '['); encode_nl (enc); 307 encode_ch (enc, '['); encode_nl (enc);
301 ++enc->indent; 308 ++enc->indent;
302 309
303 for (i = 0; i <= len; ++i) 310 for (i = 0; i <= len; ++i)
304 { 311 {
312 SV **svp = av_fetch (av, i, 0);
313
305 encode_indent (enc); 314 encode_indent (enc);
306 encode_sv (enc, *av_fetch (av, i, 0)); 315
316 if (svp)
317 encode_sv (enc, *svp);
318 else
319 encode_str (enc, "null", 4, 0);
307 320
308 if (i < len) 321 if (i < len)
309 encode_comma (enc); 322 encode_comma (enc);
310 } 323 }
311 324
334 else 347 else
335 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 348 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
336 349
337 encode_ch (enc, '"'); 350 encode_ch (enc, '"');
338 351
339 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 352 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
340 encode_ch (enc, ':'); 353 encode_ch (enc, ':');
341 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 354 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
342 encode_sv (enc, HeVAL (he)); 355 encode_sv (enc, HeVAL (he));
343} 356}
344 357
345// compare hash entries, used when all keys are bytestrings 358// compare hash entries, used when all keys are bytestrings
346static int 359static int
381 { 394 {
382 // for canonical output we have to sort by keys first 395 // for canonical output we have to sort by keys first
383 // actually, this is mostly due to the stupid so-called 396 // actually, this is mostly due to the stupid so-called
384 // security workaround added somewhere in 5.8.x. 397 // security workaround added somewhere in 5.8.x.
385 // that randomises hash orderings 398 // that randomises hash orderings
386 if (enc->flags & F_CANONICAL) 399 if (enc->json.flags & F_CANONICAL)
387 { 400 {
388 int fast = 1; 401 int fast = 1;
389 HE *he; 402 HE *he;
390#if defined(__BORLANDC__) || defined(_MSC_VER) 403#if defined(__BORLANDC__) || defined(_MSC_VER)
391 HE **hes = _alloca (count * sizeof (HE)); 404 HE **hes = _alloca (count * sizeof (HE));
463 svtype svt; 476 svtype svt;
464 477
465 SvGETMAGIC (sv); 478 SvGETMAGIC (sv);
466 svt = SvTYPE (sv); 479 svt = SvTYPE (sv);
467 480
481 if (expect_false (SvOBJECT (sv)))
482 {
483 if (SvSTASH (sv) == json_boolean_stash)
484 {
485 if (SvIV (sv))
486 encode_str (enc, "true", 4, 0);
487 else
488 encode_str (enc, "false", 5, 0);
489 }
490 else
491 {
492#if 0
493 if (0 && sv_derived_from (rv, "JSON::Literal"))
494 {
495 // not yet
496 }
497#endif
498 if (enc->json.flags & F_CONV_BLESSED)
499 {
500 // we re-bless the reference to get overload and other niceties right
501 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
502
503 if (to_json)
504 {
505 int count;
506 dSP;
507
508 ENTER; SAVETMPS; PUSHMARK (SP);
509 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
510
511 // calling with G_SCALAR ensures that we always get a 1 return value
512 PUTBACK;
513 call_sv ((SV *)GvCV (to_json), G_SCALAR);
514 SPAGAIN;
515
516 // catch this surprisingly common error
517 if (SvROK (TOPs) && SvRV (TOPs) == sv)
518 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
519
520 sv = POPs;
521 PUTBACK;
522
523 encode_sv (enc, sv);
524
525 FREETMPS; LEAVE;
526 }
527 else if (enc->json.flags & F_ALLOW_BLESSED)
528 encode_str (enc, "null", 4, 0);
529 else
530 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
531 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
532 }
533 else if (enc->json.flags & F_ALLOW_BLESSED)
534 encode_str (enc, "null", 4, 0);
535 else
536 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
537 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
538 }
539 }
468 if (svt == SVt_PVHV) 540 else if (svt == SVt_PVHV)
469 encode_hv (enc, (HV *)sv); 541 encode_hv (enc, (HV *)sv);
470 else if (svt == SVt_PVAV) 542 else if (svt == SVt_PVAV)
471 encode_av (enc, (AV *)sv); 543 encode_av (enc, (AV *)sv);
472 else if (svt < SVt_PVAV) 544 else if (svt < SVt_PVAV)
473 { 545 {
474 if (SvNIOK (sv) && SvIV (sv) == 0) 546 STRLEN len = 0;
547 char *pv = svt ? SvPV (sv, len) : 0;
548
549 if (len == 1 && *pv == '1')
550 encode_str (enc, "true", 4, 0);
551 else if (len == 1 && *pv == '0')
475 encode_str (enc, "false", 5, 0); 552 encode_str (enc, "false", 5, 0);
476 else if (SvNIOK (sv) && SvIV (sv) == 1)
477 encode_str (enc, "true", 4, 0);
478 else 553 else
479 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 554 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
480 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 555 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
481 } 556 }
482 else 557 else
497 encode_str (enc, str, len, SvUTF8 (sv)); 572 encode_str (enc, str, len, SvUTF8 (sv));
498 encode_ch (enc, '"'); 573 encode_ch (enc, '"');
499 } 574 }
500 else if (SvNOKp (sv)) 575 else if (SvNOKp (sv))
501 { 576 {
577 // trust that perl will do the right thing w.r.t. JSON syntax.
502 need (enc, NV_DIG + 32); 578 need (enc, NV_DIG + 32);
503 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 579 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
504 enc->cur += strlen (enc->cur); 580 enc->cur += strlen (enc->cur);
505 } 581 }
506 else if (SvIOKp (sv)) 582 else if (SvIOKp (sv))
507 { 583 {
508 // we assume we can always read an IV as a UV 584 // we assume we can always read an IV as a UV
509 if (SvUV (sv) & ~(UV)0x7fff) 585 if (SvUV (sv) & ~(UV)0x7fff)
510 { 586 {
587 // large integer, use the (rather slow) snprintf way.
511 need (enc, sizeof (UV) * 3); 588 need (enc, sizeof (UV) * 3);
512 enc->cur += 589 enc->cur +=
513 SvIsUV(sv) 590 SvIsUV(sv)
514 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv)) 591 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
515 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv)); 592 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
518 { 595 {
519 // optimise the "small number case" 596 // optimise the "small number case"
520 // code will likely be branchless and use only a single multiplication 597 // code will likely be branchless and use only a single multiplication
521 I32 i = SvIV (sv); 598 I32 i = SvIV (sv);
522 U32 u; 599 U32 u;
600 char digit, nz = 0;
523 601
524 need (enc, 6); 602 need (enc, 6);
525 603
526 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0; 604 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
527 u = i < 0 ? -i : i; 605 u = i < 0 ? -i : i;
528 606
529 // convert to 4.28 fixed-point representation 607 // convert to 4.28 fixed-point representation
530 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits 608 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
531 609
532 char digit, nz = 0; 610 // now output digit by digit, each time masking out the integer part
533 611 // and multiplying by 5 while moving the decimal point one to the right,
612 // resulting in a net multiplication by 10.
613 // we always write the digit to memory but conditionally increment
614 // the pointer, to ease the usage of conditional move instructions.
534 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5; 615 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
535 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5; 616 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
536 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5; 617 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
537 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5; 618 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
538 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; 619 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
539 } 620 }
540 } 621 }
541 else if (SvROK (sv)) 622 else if (SvROK (sv))
542 encode_rv (enc, SvRV (sv)); 623 encode_rv (enc, SvRV (sv));
543 else if (!SvOK (sv)) 624 else if (!SvOK (sv))
546 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 627 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
547 SvPV_nolen (sv), SvFLAGS (sv)); 628 SvPV_nolen (sv), SvFLAGS (sv));
548} 629}
549 630
550static SV * 631static SV *
551encode_json (SV *scalar, U32 flags) 632encode_json (SV *scalar, JSON *json)
552{ 633{
553 enc_t enc; 634 enc_t enc;
554 635
555 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 636 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
556 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 637 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
557 638
558 enc.flags = flags; 639 enc.json = *json;
559 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 640 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
560 enc.cur = SvPVX (enc.sv); 641 enc.cur = SvPVX (enc.sv);
561 enc.end = SvEND (enc.sv); 642 enc.end = SvEND (enc.sv);
562 enc.indent = 0; 643 enc.indent = 0;
563 enc.maxdepth = DEC_DEPTH (flags); 644 enc.maxdepth = DEC_DEPTH (enc.json.flags);
564 645
565 SvPOK_only (enc.sv); 646 SvPOK_only (enc.sv);
566 encode_sv (&enc, scalar); 647 encode_sv (&enc, scalar);
567 648
568 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 649 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
569 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 650 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
570 651
571 if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8))) 652 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
572 SvUTF8_on (enc.sv); 653 SvUTF8_on (enc.sv);
573 654
574 if (enc.flags & F_SHRINK) 655 if (enc.json.flags & F_SHRINK)
575 shrink (enc.sv); 656 shrink (enc.sv);
576 657
577 return enc.sv; 658 return enc.sv;
578} 659}
579 660
584typedef struct 665typedef struct
585{ 666{
586 char *cur; // current parser pointer 667 char *cur; // current parser pointer
587 char *end; // end of input string 668 char *end; // end of input string
588 const char *err; // parse error, if != 0 669 const char *err; // parse error, if != 0
589 U32 flags; // F_* 670 JSON json;
590 U32 depth; // recursion depth 671 U32 depth; // recursion depth
591 U32 maxdepth; // recursion depth limit 672 U32 maxdepth; // recursion depth limit
592} dec_t; 673} dec_t;
593 674
594inline void 675inline void
625decode_4hex (dec_t *dec) 706decode_4hex (dec_t *dec)
626{ 707{
627 signed char d1, d2, d3, d4; 708 signed char d1, d2, d3, d4;
628 unsigned char *cur = (unsigned char *)dec->cur; 709 unsigned char *cur = (unsigned char *)dec->cur;
629 710
630 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("four hexadecimal digits expected"); 711 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
631 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("four hexadecimal digits expected"); 712 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
632 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("four hexadecimal digits expected"); 713 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
633 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("four hexadecimal digits expected"); 714 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
634 715
635 dec->cur += 4; 716 dec->cur += 4;
636 717
637 return ((UV)d1) << 12 718 return ((UV)d1) << 12
638 | ((UV)d2) << 8 719 | ((UV)d2) << 8
857 is_nv = 1; 938 is_nv = 1;
858 } 939 }
859 940
860 if (!is_nv) 941 if (!is_nv)
861 { 942 {
943 int len = dec->cur - start;
944
862 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so 945 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
863 if (*start == '-') 946 if (*start == '-')
864 switch (dec->cur - start) 947 switch (len)
865 { 948 {
866 case 2: return newSViv (-( start [1] - '0' )); 949 case 2: return newSViv (-( start [1] - '0' * 1));
867 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 950 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
868 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); 951 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
869 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); 952 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
870 } 953 }
871 else 954 else
872 switch (dec->cur - start) 955 switch (len)
873 { 956 {
874 case 1: return newSViv ( start [0] - '0' ); 957 case 1: return newSViv ( start [0] - '0' * 1);
875 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 958 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
876 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 959 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
877 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 960 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
878 } 961 }
879 962
880 { 963 {
881 UV uv; 964 UV uv;
882 int numtype = grok_number (start, dec->cur - start, &uv); 965 int numtype = grok_number (start, len, &uv);
883 if (numtype & IS_NUMBER_IN_UV) 966 if (numtype & IS_NUMBER_IN_UV)
884 if (numtype & IS_NUMBER_NEG) 967 if (numtype & IS_NUMBER_NEG)
885 { 968 {
886 if (uv < (UV)IV_MIN) 969 if (uv < (UV)IV_MIN)
887 return newSViv (-(IV)uv); 970 return newSViv (-(IV)uv);
888 } 971 }
889 else 972 else
890 return newSVuv (uv); 973 return newSVuv (uv);
891 } 974 }
892 }
893 975
976 len -= *start == '-' ? 1 : 0;
977
978 // does not fit into IV or UV, try NV
979 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len)
980 #if defined (LDBL_DIG)
981 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
982 #endif
983 )
984 // fits into NV without loss of precision
985 return newSVnv (Atof (start));
986
987 // everything else fails, convert it to a string
988 return newSVpvn (start, dec->cur - start);
989 }
990
991 // loss of precision here
894 return newSVnv (Atof (start)); 992 return newSVnv (Atof (start));
895 993
896fail: 994fail:
897 return 0; 995 return 0;
898} 996}
942} 1040}
943 1041
944static SV * 1042static SV *
945decode_hv (dec_t *dec) 1043decode_hv (dec_t *dec)
946{ 1044{
1045 SV *sv;
947 HV *hv = newHV (); 1046 HV *hv = newHV ();
948 1047
949 DEC_INC_DEPTH; 1048 DEC_INC_DEPTH;
950 decode_ws (dec); 1049 decode_ws (dec);
951 1050
952 if (*dec->cur == '}') 1051 if (*dec->cur == '}')
953 ++dec->cur; 1052 ++dec->cur;
954 else 1053 else
955 for (;;) 1054 for (;;)
956 { 1055 {
957 SV *key, *value;
958
959 decode_ws (dec); EXPECT_CH ('"'); 1056 decode_ws (dec); EXPECT_CH ('"');
960 1057
961 key = decode_str (dec); 1058 // heuristic: assume that
962 if (!key) 1059 // a) decode_str + hv_store_ent are abysmally slow.
963 goto fail; 1060 // b) most hash keys are short, simple ascii text.
1061 // => try to "fast-match" such strings to avoid
1062 // the overhead of decode_str + hv_store_ent.
1063 {
1064 SV *value;
1065 char *p = dec->cur;
1066 char *e = p + 24; // only try up to 24 bytes
964 1067
965 decode_ws (dec); EXPECT_CH (':'); 1068 for (;;)
966
967 value = decode_sv (dec);
968 if (!value)
969 { 1069 {
1070 // the >= 0x80 is true on most architectures
1071 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1072 {
1073 // slow path, back up and use decode_str
1074 SV *key = decode_str (dec);
1075 if (!key)
1076 goto fail;
1077
1078 decode_ws (dec); EXPECT_CH (':');
1079
1080 value = decode_sv (dec);
1081 if (!value)
1082 {
1083 SvREFCNT_dec (key);
1084 goto fail;
1085 }
1086
1087 hv_store_ent (hv, key, value, 0);
970 SvREFCNT_dec (key); 1088 SvREFCNT_dec (key);
1089
1090 break;
1091 }
1092 else if (*p == '"')
1093 {
1094 // fast path, got a simple key
1095 char *key = dec->cur;
1096 int len = p - key;
1097 dec->cur = p + 1;
1098
1099 decode_ws (dec); EXPECT_CH (':');
1100
1101 value = decode_sv (dec);
1102 if (!value)
971 goto fail; 1103 goto fail;
1104
1105 hv_store (hv, key, len, value, 0);
1106
1107 break;
1108 }
1109
1110 ++p;
972 } 1111 }
973 1112 }
974 hv_store_ent (hv, key, value, 0);
975 SvREFCNT_dec (key);
976 1113
977 decode_ws (dec); 1114 decode_ws (dec);
978 1115
979 if (*dec->cur == '}') 1116 if (*dec->cur == '}')
980 { 1117 {
987 1124
988 ++dec->cur; 1125 ++dec->cur;
989 } 1126 }
990 1127
991 DEC_DEC_DEPTH; 1128 DEC_DEC_DEPTH;
992 return newRV_noinc ((SV *)hv); 1129 sv = newRV_noinc ((SV *)hv);
1130
1131 // check filter callbacks
1132 if (dec->json.flags & F_HOOK)
1133 {
1134 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1135 {
1136 HE *cb, *he;
1137
1138 hv_iterinit (hv);
1139 he = hv_iternext (hv);
1140 hv_iterinit (hv);
1141
1142 // the next line creates a mortal sv each time its called.
1143 // might want to optimise this for common cases.
1144 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1145
1146 if (cb)
1147 {
1148 dSP;
1149 int count;
1150
1151 ENTER; SAVETMPS; PUSHMARK (SP);
1152 XPUSHs (HeVAL (he));
1153
1154 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1155
1156 if (count == 1)
1157 {
1158 sv = newSVsv (POPs);
1159 FREETMPS; LEAVE;
1160 return sv;
1161 }
1162
1163 FREETMPS; LEAVE;
1164 }
1165 }
1166
1167 if (dec->json.cb_object)
1168 {
1169 dSP;
1170 int count;
1171
1172 ENTER; SAVETMPS; PUSHMARK (SP);
1173 XPUSHs (sv_2mortal (sv));
1174
1175 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1176
1177 if (count == 1)
1178 {
1179 sv = newSVsv (POPs);
1180 FREETMPS; LEAVE;
1181 return sv;
1182 }
1183
1184 SvREFCNT_inc (sv);
1185 FREETMPS; LEAVE;
1186 }
1187 }
1188
1189 return sv;
993 1190
994fail: 1191fail:
995 SvREFCNT_dec (hv); 1192 SvREFCNT_dec (hv);
996 DEC_DEC_DEPTH; 1193 DEC_DEC_DEPTH;
997 return 0; 1194 return 0;
999 1196
1000static SV * 1197static SV *
1001decode_sv (dec_t *dec) 1198decode_sv (dec_t *dec)
1002{ 1199{
1003 decode_ws (dec); 1200 decode_ws (dec);
1201
1202 // the beauty of JSON: you need exactly one character lookahead
1203 // to parse anything.
1004 switch (*dec->cur) 1204 switch (*dec->cur)
1005 { 1205 {
1006 case '"': ++dec->cur; return decode_str (dec); 1206 case '"': ++dec->cur; return decode_str (dec);
1007 case '[': ++dec->cur; return decode_av (dec); 1207 case '[': ++dec->cur; return decode_av (dec);
1008 case '{': ++dec->cur; return decode_hv (dec); 1208 case '{': ++dec->cur; return decode_hv (dec);
1014 1214
1015 case 't': 1215 case 't':
1016 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1216 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1017 { 1217 {
1018 dec->cur += 4; 1218 dec->cur += 4;
1019 return newSViv (1); 1219 return SvREFCNT_inc (json_true);
1020 } 1220 }
1021 else 1221 else
1022 ERR ("'true' expected"); 1222 ERR ("'true' expected");
1023 1223
1024 break; 1224 break;
1025 1225
1026 case 'f': 1226 case 'f':
1027 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1227 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1028 { 1228 {
1029 dec->cur += 5; 1229 dec->cur += 5;
1030 return newSViv (0); 1230 return SvREFCNT_inc (json_false);
1031 } 1231 }
1032 else 1232 else
1033 ERR ("'false' expected"); 1233 ERR ("'false' expected");
1034 1234
1035 break; 1235 break;
1053fail: 1253fail:
1054 return 0; 1254 return 0;
1055} 1255}
1056 1256
1057static SV * 1257static SV *
1058decode_json (SV *string, U32 flags, UV *offset_return) 1258decode_json (SV *string, JSON *json, UV *offset_return)
1059{ 1259{
1060 dec_t dec; 1260 dec_t dec;
1061 UV offset; 1261 UV offset;
1062 SV *sv; 1262 SV *sv;
1063 1263
1064 SvGETMAGIC (string); 1264 SvGETMAGIC (string);
1065 SvUPGRADE (string, SVt_PV); 1265 SvUPGRADE (string, SVt_PV);
1066 1266
1267 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
1268 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1269 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
1270
1067 if (flags & F_UTF8) 1271 if (json->flags & F_UTF8)
1068 sv_utf8_downgrade (string, 0); 1272 sv_utf8_downgrade (string, 0);
1069 else 1273 else
1070 sv_utf8_upgrade (string); 1274 sv_utf8_upgrade (string);
1071 1275
1072 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1276 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1073 1277
1074 dec.flags = flags; 1278 dec.json = *json;
1075 dec.cur = SvPVX (string); 1279 dec.cur = SvPVX (string);
1076 dec.end = SvEND (string); 1280 dec.end = SvEND (string);
1077 dec.err = 0; 1281 dec.err = 0;
1078 dec.depth = 0; 1282 dec.depth = 0;
1079 dec.maxdepth = DEC_DEPTH (dec.flags); 1283 dec.maxdepth = DEC_DEPTH (dec.json.flags);
1284
1285 if (dec.json.cb_object || dec.json.cb_sk_object)
1286 dec.json.flags |= F_HOOK;
1080 1287
1081 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1288 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1082 sv = decode_sv (&dec); 1289 sv = decode_sv (&dec);
1083 1290
1084 if (!(offset_return || !sv)) 1291 if (!(offset_return || !sv))
1094 } 1301 }
1095 } 1302 }
1096 1303
1097 if (offset_return || !sv) 1304 if (offset_return || !sv)
1098 { 1305 {
1099 offset = dec.flags & F_UTF8 1306 offset = dec.json.flags & F_UTF8
1100 ? dec.cur - SvPVX (string) 1307 ? dec.cur - SvPVX (string)
1101 : utf8_distance (dec.cur, SvPVX (string)); 1308 : utf8_distance (dec.cur, SvPVX (string));
1102 1309
1103 if (offset_return) 1310 if (offset_return)
1104 *offset_return = offset; 1311 *offset_return = offset;
1123 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1330 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1124 } 1331 }
1125 1332
1126 sv = sv_2mortal (sv); 1333 sv = sv_2mortal (sv);
1127 1334
1128 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1335 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1129 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1336 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1130 1337
1131 return sv; 1338 return sv;
1132} 1339}
1133 1340
1137MODULE = JSON::XS PACKAGE = JSON::XS 1344MODULE = JSON::XS PACKAGE = JSON::XS
1138 1345
1139BOOT: 1346BOOT:
1140{ 1347{
1141 int i; 1348 int i;
1142
1143 memset (decode_hexdigit, 0xff, 256);
1144 1349
1145 for (i = 0; i < 256; ++i) 1350 for (i = 0; i < 256; ++i)
1146 decode_hexdigit [i] = 1351 decode_hexdigit [i] =
1147 i >= '0' && i <= '9' ? i - '0' 1352 i >= '0' && i <= '9' ? i - '0'
1148 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1353 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1149 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1354 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1150 : -1; 1355 : -1;
1151 1356
1152 json_stash = gv_stashpv ("JSON::XS", 1); 1357 json_stash = gv_stashpv ("JSON::XS" , 1);
1358 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1359
1360 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
1361 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1153} 1362}
1154 1363
1155PROTOTYPES: DISABLE 1364PROTOTYPES: DISABLE
1156 1365
1157SV *new (char *dummy) 1366void new (char *klass)
1158 CODE: 1367 PPCODE:
1159 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1368{
1160 OUTPUT: 1369 SV *pv = NEWSV (0, sizeof (JSON));
1161 RETVAL 1370 SvPOK_only (pv);
1371 Zero (SvPVX (pv), 1, JSON);
1372 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1373 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), json_stash)));
1374}
1162 1375
1163SV *ascii (SV *self, int enable = 1) 1376void ascii (JSON *self, int enable = 1)
1164 ALIAS: 1377 ALIAS:
1165 ascii = F_ASCII 1378 ascii = F_ASCII
1166 latin1 = F_LATIN1 1379 latin1 = F_LATIN1
1167 utf8 = F_UTF8 1380 utf8 = F_UTF8
1168 indent = F_INDENT 1381 indent = F_INDENT
1169 canonical = F_CANONICAL 1382 canonical = F_CANONICAL
1170 space_before = F_SPACE_BEFORE 1383 space_before = F_SPACE_BEFORE
1171 space_after = F_SPACE_AFTER 1384 space_after = F_SPACE_AFTER
1172 pretty = F_PRETTY 1385 pretty = F_PRETTY
1173 allow_nonref = F_ALLOW_NONREF 1386 allow_nonref = F_ALLOW_NONREF
1174 shrink = F_SHRINK 1387 shrink = F_SHRINK
1388 allow_blessed = F_ALLOW_BLESSED
1389 convert_blessed = F_CONV_BLESSED
1175 CODE: 1390 PPCODE:
1176{ 1391{
1177 UV *uv = SvJSON (self);
1178 if (enable) 1392 if (enable)
1179 *uv |= ix; 1393 self->flags |= ix;
1180 else 1394 else
1181 *uv &= ~ix; 1395 self->flags &= ~ix;
1182 1396
1183 RETVAL = newSVsv (self); 1397 XPUSHs (ST (0));
1184} 1398}
1185 OUTPUT:
1186 RETVAL
1187 1399
1188SV *max_depth (SV *self, UV max_depth = 0x80000000UL) 1400void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1189 CODE: 1401 PPCODE:
1190{ 1402{
1191 UV *uv = SvJSON (self);
1192 UV log2 = 0; 1403 UV log2 = 0;
1193 1404
1194 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 1405 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1195 1406
1196 while ((1UL << log2) < max_depth) 1407 while ((1UL << log2) < max_depth)
1197 ++log2; 1408 ++log2;
1198 1409
1199 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH); 1410 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1200 1411
1201 RETVAL = newSVsv (self); 1412 XPUSHs (ST (0));
1202} 1413}
1203 OUTPUT:
1204 RETVAL
1205 1414
1206void encode (SV *self, SV *scalar) 1415void max_size (JSON *self, UV max_size = 0)
1207 PPCODE: 1416 PPCODE:
1208 XPUSHs (encode_json (scalar, *SvJSON (self))); 1417{
1418 UV log2 = 0;
1209 1419
1210void decode (SV *self, SV *jsonstr) 1420 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1421 if (max_size == 1) max_size = 2;
1422
1423 while ((1UL << log2) < max_size)
1424 ++log2;
1425
1426 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1427
1428 XPUSHs (ST (0));
1429}
1430
1431void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1211 PPCODE: 1432 PPCODE:
1433{
1434 SvREFCNT_dec (self->cb_object);
1435 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1436
1437 XPUSHs (ST (0));
1438}
1439
1440void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1441 PPCODE:
1442{
1443 if (!self->cb_sk_object)
1444 self->cb_sk_object = newHV ();
1445
1446 if (SvOK (cb))
1447 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1448 else
1449 {
1450 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1451
1452 if (!HvKEYS (self->cb_sk_object))
1453 {
1454 SvREFCNT_dec (self->cb_sk_object);
1455 self->cb_sk_object = 0;
1456 }
1457 }
1458
1459 XPUSHs (ST (0));
1460}
1461
1462void encode (JSON *self, SV *scalar)
1463 PPCODE:
1464 XPUSHs (encode_json (scalar, self));
1465
1466void decode (JSON *self, SV *jsonstr)
1467 PPCODE:
1212 XPUSHs (decode_json (jsonstr, *SvJSON (self), 0)); 1468 XPUSHs (decode_json (jsonstr, self, 0));
1213 1469
1214void decode_prefix (SV *self, SV *jsonstr) 1470void decode_prefix (JSON *self, SV *jsonstr)
1215 PPCODE: 1471 PPCODE:
1216{ 1472{
1217 UV offset; 1473 UV offset;
1218 EXTEND (SP, 2); 1474 EXTEND (SP, 2);
1219 PUSHs (decode_json (jsonstr, *SvJSON (self), &offset)); 1475 PUSHs (decode_json (jsonstr, self, &offset));
1220 PUSHs (sv_2mortal (newSVuv (offset))); 1476 PUSHs (sv_2mortal (newSVuv (offset)));
1221} 1477}
1222 1478
1479void DESTROY (JSON *self)
1480 CODE:
1481 SvREFCNT_dec (self->cb_sk_object);
1482 SvREFCNT_dec (self->cb_object);
1483
1223PROTOTYPES: ENABLE 1484PROTOTYPES: ENABLE
1224 1485
1225void to_json (SV *scalar) 1486void to_json (SV *scalar)
1226 ALIAS:
1227 objToJson = 0
1228 PPCODE: 1487 PPCODE:
1488{
1489 JSON json = { F_DEFAULT | F_UTF8 };
1229 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8)); 1490 XPUSHs (encode_json (scalar, &json));
1491}
1230 1492
1231void from_json (SV *jsonstr) 1493void from_json (SV *jsonstr)
1232 ALIAS:
1233 jsonToObj = 0
1234 PPCODE: 1494 PPCODE:
1495{
1496 JSON json = { F_DEFAULT | F_UTF8 };
1235 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0)); 1497 XPUSHs (decode_json (jsonstr, &json, 0));
1498}
1236 1499

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines